A scaly artificial neural network architecture for isolated word recognition using the British Telecommunications English alphabet database

The use of neural networks for the recognition of isolated letters from the English alphabet is investigated. A scaly architecture neural network model is used and is trained using the error backpropagation algorithm. Two different methods of nonlinear time alignment are used and the performance of the neural network when used with each is compared. The use of different transfer functions within the processing units of the neural network is examined, and several versions of the sigmoid function are compared. The variance of the learning rate in the error backpropagation algorithm is investigated by comparing the performance of the neural network when three different values of learning rate are used